Headship Rates and Housing Demand Michael Carliner The strength of housing demand in recent years is related to an increase in the rate of net household formations. From March 1990 to March 1996, the average increase in the number of households was about 1.04 million per year, according to the Census Bureau s Current Population Survey (CPS). From March 1996 to March 2002, the average increase was 1.31 million. That means that the number of housing units absorbed per year for use as primary residences was 270,000 greater since 1996. 1 Although growth in the adult population and changes in the population age structure accounted for most of the overall increase in households, the acceleration in the late 1990s was largely attributable to a greater tendency for young adults to establish their own households, rather than live with parents, other relatives, or friends. Those shifts in behavior produced increases in headship rates (the percentage of all people in an age group or other category who are household heads). 2 Among the population aged 25 to 34, for example, the share who were household heads increased after 1996 by more than two and a half percentage points. With about 40 million people in that age group, that meant substantially more households. The (somewhat belated) tendency for the Generation X population to establish independent households, to a greater extent than the late-baby Boomers that preceded them, together with a wave of immigration that has swollen the Gen-X ranks, has meant a more balanced mix of housing demand than was earlier anticipated. There has been no glut of starter homes as a result of the aging of the Baby Boom. Table 1. Households and Headship Rates By Age 1980 weights 1990 weights 2000 weights Change 1990 1992 1993 1993 1994 1996 1998 2000 2001 2001 2002 90-96 96-02 92-02 Population (non-institutional) Total 245.5 251.6 252.5 253.6 258.3 264.3 269.1 274.1 276.8 279.5 282.1 18.8 12.5 25.2 15 to 24 35.3 34.4 34.4 36.2 36.3 36.3 37.1 38.5 39.0 38.8 39.4 1.0 2.7 4.6 25 to 34 43.2 42.5 41.9 42.4 41.9 40.9 39.4 37.8 37.4 38.9 38.7 (2.3) (3.5) (5.1) 35 to 44 37.2 39.6 40.3 40.7 41.5 43.1 44.5 44.8 44.8 44.6 44.3 5.9 1.7 5.2 45 to 54 25.3 27.0 28.5 28.3 29.5 31.6 34.1 36.6 38.0 38.7 39.5 6.3 6.5 11.0 55 to 64 21.2 21.2 21.2 20.5 20.7 21.1 22.3 23.4 23.8 24.7 25.9 (0.1) 2.7 2.6 65 and over 29.6 30.6 30.9 30.4 30.8 31.7 32.1 32.6 33.0 33.6 33.8 2.1 1.3 2.4 15 and over 191.8 195.2 197.3 198.6 200.8 204.6 209.3 213.8 216.0 219.2 221.6 12.8 11.4 20.8 Households Total 93.3 95.7 96.4 96.4 97.1 99.6 102.5 104.7 106.4 108.2 109.3 6.2 7.9 11.8 15 to 24 5.1 4.9 5.0 5.3 5.3 5.3 5.4 5.9 6.4 6.4 6.4 (0.1) 1.1 1.3 25 to 34 20.5 20.0 19.7 20.1 19.7 19.2 19.0 18.6 18.6 19.0 19.0 (1.6) (0.7) (1.8) 35 to 44 20.6 21.8 21.7 21.9 22.3 23.2 23.9 24.0 23.9 24.1 24.0 2.5 0.7 2.0 45 to 54 14.5 15.5 16.6 16.4 16.8 18.0 19.5 20.9 21.8 22.0 22.2 3.7 4.0 6.7 55 to 64 12.5 12.6 12.4 12.2 12.2 12.4 13.1 13.6 13.9 14.3 15.2 0.2 2.5 2.6 65 and over 20.2 20.9 20.9 20.7 20.8 21.5 21.5 21.7 21.8 22.5 22.5 1.5 0.3 1.1 Headship Rate 15 to 24 14.5 14.1 14.6 14.5 14.5 14.5 14.7 15.2 16.4 16.5 16.2 0.1 1.5 2.0 25 to 34 47.3 47.1 47.2 47.4 47.0 47.0 48.4 49.3 49.6 49.0 49.1-0.6 2.7 2.4 35 to 44 55.3 55.0 53.8 53.7 53.7 53.9 53.9 53.5 53.4 54.0 54.3-1.2-0.2-1.2 45 to 54 57.4 57.5 58.2 57.9 57.0 57.0 57.4 57.1 57.3 56.7 56.2-0.1-0.3-0.6 55 to 64 59.0 59.4 58.5 59.2 58.8 58.8 58.8 58.1 58.6 57.9 58.8-0.9 0.7-0.5 65 and over 68.2 68.4 67.7 68.0 67.6 67.9 67.0 66.7 66.2 66.9 66.6-0.6-2.1-2.9 Adjusted Headship Rate* 15 to 24 14.3 13.8 14.3 14.2 14.1 14.0 14.0 14.5 15.4 15.6 15.2-0.2 1.0 1.3 25 to 34 47.2 46.8 46.8 47.0 46.7 46.4 47.5 48.2 48.6 47.8 47.8-1.0 2.2 1.6 35 to 44 55.2 55.0 53.8 53.6 53.5 53.9 53.6 53.4 53.0 53.8 54.2-1.1-0.5-1.5 45 to 54 57.5 57.5 58.2 58.0 57.3 57.1 57.8 57.2 57.6 57.0 56.4-0.2 0.0-0.3 55 to 64 59.1 59.7 59.0 59.7 59.1 59.6 59.2 59.0 59.3 58.6 59.5-0.2 0.6-0.3 65 and over 68.6 68.8 68.2 68.4 68.3 68.8 68.5 68.2 68.1 68.8 68.5-0.1-1.0-1.3 Notes: Population and Households are in millions. *Adjusted to set male in married couple as head of household. 8 June 2003
Population Data The CPS estimates of the number of households are based on independent estimates of the population by age, race, Hispanic origin, and sex. The population estimates are largely extrapolations from the previous decennial census. The net growth in Table 1 is calculated by comparing estimates of the number of households from successive years using comparable population estimates, removing the large adjustments that occur when the base is changed every ten years or so. This makes it easier to analyze trends in headship rates, but because population actually grew faster than the estimates, the average growth in the number of households is understated. From 1990 to 2000, the average annual increase in the number of households, based on the March CPS, was 1.13 million per year. The 2000 Census, however, counted 13.5 million more households than the 1990 Census, indicating average net growth of 1.35 million per year. The population estimates used for the CPS were based on assumptions of net immigration averaging about 870,000 per year. That now appears to have been too low, by perhaps 200,000 persons per year. The growth indicated by the decennial census counts in 1990 and 2000, however, is an overstatement. The Census Bureau s revised estimates of the accuracy of the decennial census, reported in March 2003, indicate that population was overcounted by about 1.3 million in 2000, compared to an undercount of about 4.0 million in 1990. 3 Much of the shift from undercount to overcount involved children under 18, but the adult population coverage changed as well. The recent report on accuracy of the 2000 census does not include estimates of the accuracy of the count of households or of vacancies, but the difference between the 2000 population overcount and the 1990 undercount for various age groups probably means that the actual annual average increase in the number of households from 1990 to 2000 was around 1.24 million. For the period since the April 1, 2000, the Census Bureau recently released estimates of population by age incorporating assumptions of net immigration of about 1.5 million per year, based on a survey of 700,000 households conducted in 2001. Even if the estimate of net immigration of 1.8 million for the period from April 2000 to July 2001 was correct, it is difficult to imagine that net immigration was not reduced by recession and homeland security measures. Earlier post-2000 population estimates were more conservative than the new figures, and the higher figures were not used as controls for the CPS until January 2003. Headship Rate Change After adjusting for reweighting of the Current Population Survey in 1993 and 2001, the headship rates for the population 15 to 24 and 25 to 34 increased by 2.0 and 2.4 percentage 63 58 53 48 43 points, respectively, over the decade between 1992 and 2002, with most of those increases occurring after 1996. Headship rates among older age groups fell, however. For the 35 to 44 age bracket, the decline from 1992 to 2002 was 1.2 percentage points, with more modest declines of 0.6 percent and 0.5 percent among those 45 to 54 and 55 to 64, respectively. The largest decline was shown for the population 65 and over, for which the headship rate fell by 2.9 percentage points. If headship rates by age had remained at their 1992 values, changes in population by age, as used in the CPS, would have added an average of 1.20 million households per year from 1992 to 2002, but the actual average increase was 1.18 million. During the period from 1996 to 2002, the average 1.31 million increase in the number of households exceeded the 1.17 million implied by population changes with constant headship rates, but that surge was not enough to fully offset the earlier decline in headship. From 1990 to 1996 the actual average increase was only 1.04 million per year, based on the CPS, while the population change implied 1.23 million. 4 Part of the increases in headship Figure 1. Age 18 to 24 Living with Parents by Gender Percent 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 Male Female June 2003 9
17 15 13 11 9 7 Figure 2. Age 25 to 34 Living with Parents by Gender Percent 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 for the younger groups and declines for the older groups in recent years reflected the arbitrary designation of the householder within each household. Among married couple households, an increasing share had the wife recorded as the householder. In 2002, the wife was recorded as the householder for 33 percent of married-couple households, compared to only 8 percent in 1992. Since wives are younger on average than their husbands, that translated into more young householders and fewer older householders. If headship rates are calculated using the politically-incorrect and often-inaccurate convention followed until 1980 always showing the husband as the head in a married-couple household the decline from 1992 to 2002 for the oldest group is reduced to 1.3 percentage points, and the increases for the 2 youngest groups become 1.3 percent and 1.6 percent, respectively. Even after that adjustment, however, headship rates for each age group over age 34 still declined. Indeed, for the 35 to 44 year old category, the decline between 1992 and 2002 appears greater, becoming 1.5 percentage points. Male Female Household Types Households maintained by married couples are still the most common variety, but their share of total households continued to decline. In 2002, about 57 million households were maintained by married couples, representing 52 percent of all households, compared to 53 percent in 1997. Despite the decline in the married couple share of total households, the average net increase in married couple households was much greater in the last five or six years than in the early and mid 1990s. The declines in the share of households maintained by married couples have been partly offset by increases averaging about 200,000 per year in the number of unmarried couples maintaining households. The 2000 Census found about 4.9 million opposite-sex unmarriedcouple households, as well as another 0.6 million same-sex couples. 5 Since 1997, growth in the number of households consisting of people living alone also accelerated, but there was reduced growth in the number of single-parent households and other types of family households not maintained by married couples, compared to the rapid increases experienced in the early 1990s, as well as in the 1970s and 1980s. The slowdown in growth of the total number of households in the early 1990s, and the acceleration in more recent years, was largely driven by changes in the behavior of young adults. Specifically, in the early 1990s many young adults, especially young men, remained in their parents homes rather than establishing households of their own. The share of people aged 15 to 24 living with their parents increased from 63.6 percent in 1989 to 65.5 percent in 1997, but fell by more than 2 percentage points after 1997. The increase in the share of young adults living with their parents in the 1980s and in the early part of the last decade was partly attributable to postponement of marriage, since young adults who have never been married are far more likely to live with their parents than those who are divorced or separated, and only a tiny share of currently-married young adults live together in the home of a parent. The decline in the share of young adults living with their parents since the mid-1990s was not due to any increase in the share who were married. The biggest changes were in the shares of never-married young adults living with their parents, which fell sharply. For example, the decline in the population aged 25 to 34 living with their parents from 12.3 percent in 1996 to 10.2 percent in 2001 was primarily due to a decline from 29.0 percent to 22.8 percent in the proportion of the never-married population in that age group living with parents. 6 Foreign-Born U.S. household patterns are being influence by the increasing foreignborn population. Overall, about 13.7 percent of the household population 10 June 2003
aged 15 and over in March 2002 was foreign-born, but among those aged 25 to 34, the share was 19.4 percent, and the foreign-born share among those 35 to 44 was 15.8 percent. The higher shares in those age groups reflected high rates of immigration in the 1980s and 1990s, and the fact that immigrants tend to arrive as young adults. In addition, low U.S. births in the late 1960s and 1970s meant a smaller native-born population aged 25 to 34 in 2002. In most foreign countries, it is more common than in the U.S. for unmarried young adults, and even married couples, to live with their parents. Young foreign-born adults in the U.S., however, are less likely than native-born young adults to live in a household headed by a parent, in part because many of their parents are still in Mexico, China, or wherever they came from. Only 7.7 percent of the foreign-born population aged 25 to 34 lived in a parent s home in March 2002, compared to 11.7 percent of the native-born population in that age group. In the 35 to 44 age bracket, the corresponding shares were 2.6 percent and 5.1 percent, respectively. When native-born young or middle-aged adults live in the same household as their parents, it is generally the parents who are the householders, providing homes for the children. When elderly foreign-born parents live in the same household as their adult children, it is generally the (foreign- or native-born) children who are the householders, providing homes for the parents. Among the native-born population aged 35 and over in 2002, there were 3.3 million people living in households maintained by their parents, compared to 2.0 million living in households where their children were the householders. Among the foreign-born population aged 35 and over, there were only 0.3 million living in households where their parents were the householders, compared to 1.1 million living in households headed by their children. Table 2 shows the living arrangements of the native-born and foreignborn populations in different age groups. Although foreign-born young adults are not very likely to live in a parent s household, they are far more likely to live in households maintained by siblings or other relatives than their native-born counterparts. Relatively-few foreign-born adults live alone, especially if they are female. The tendency of the foreign-born population to live in extended families is greater for recent arrivals than for long-term residents, who tend to have living arrangements more similar to the native-born population. For the foreign-born population, the number of married-couple householders in March 2002 was 7.7 million, while the number who were the spouse of the householder in a married-couple household was 8.3 million. That is not because of polygamy. In cases where a foreign-born person was married to a native-born person, the native-born person was recorded as the householder in the majority of cases. Although some of the cases involved a native-born wife being counted as the householder, rather than a foreign-born husband, this phenomenon was largely due to the fact that there are somewhat more married couples in the US where the husband is native-born and the wife is foreignborn, rather than vice-versa. The effect on headship rates for the large native-born population was small, but the overall headship rate among the foreign-born population was reduced by about one percentage point, compared to what it would be if half of such couples had the foreign-born partner counted as the householder. Among the entire foreign-born population in 2002, about 49 percent had arrived in the U.S. after 1989, but among those 25 to 34 years old, the share who had arrived in 1990 or later Table 2. Living Arrangements of Native-born (NB) and Foreign-born (FB) Population 15-24 25-34 35-44 45-54 55-64 65 and Over Total (15 & up) March 2002 NB FB NB FB NB FB NB FB NB FB NB FB NB FB Population 34,902 4,547 31,179 7,491 37,279 7,005 34,614 4,931 22,820 3,054 30,460 3,309191,254 30,337 Householders 15.6% 20.9% 50.8% 42.0% 55.2% 49.3% 56.9% 51.1% 59.8% 50.8% 67.9% 54.1% 50.1% 44.2% Married Couple 3.6% 6.1% 24.4% 23.6% 31.8% 33.2% 33.1% 33.8% 35.1% 29.7% 28.9% 23.8% 25.6% 25.5% Other Family 5.0% 7.5% 10.2% 8.8% 10.7% 9.3% 8.5% 9.2% 5.9% 7.7% 6.0% 6.6% 7.9% 8.4% Live Alone 3.3% 3.3% 11.0% 6.4% 10.0% 5.2% 12.8% 6.6% 17.1% 11.4% 31.9% 22.4% 13.8% 7.9% Other Nonfamily 3.7% 4.1% 5.1% 3.3% 2.7% 1.5% 2.5% 1.5% 1.7% 1.9% 1.1% 1.3% 2.9% 2.4% Non-Householders 84.4% 78.9% 49.1% 57.9% 44.7% 50.7% 43.0% 48.8% 40.1% 49.2% 32.0% 45.9% 49.8% 55.7% Spouse 4.1% 7.3% 25.6% 28.0% 32.8% 35.3% 33.2% 33.5% 33.4% 31.5% 25.3% 23.7% 25.3% 27.4% Child 65.2% 41.2% 11.7% 7.7% 5.1% 2.6% 3.4% 1.1% 1.4% 0.5% 0.1% 0.2% 15.6% 8.9% Parent 0.0% 0.1% 0.1% 0.1% 0.5% 1.7% 1.5% 5.9% 1.5% 9.2% 3.3% 12.7% 1.1% 3.7% Other Relative 6.3% 18.2% 2.6% 11.2% 1.6% 5.4% 1.7% 3.6% 1.2% 5.3% 2.1% 7.9% 2.7% 8.7% Non-relative 8.8% 12.1% 9.2% 11.0% 4.7% 5.7% 3.3% 4.8% 2.6% 2.7% 1.2% 1.3% 5.1% 7.0% Source: U.S. Bureau of the Census, Current Population Survey, Annual Demographic Supplement 2002; tabulations by NAHB. Note: Excludes institutional population. June 2003 11
was nearly two-thirds. Yet despite their relatively-recent arrivals, the tendency of new arrivals to live in extended families, and the especially-large foreign-born share among the U.S. population aged 25 to 34 in 2002, from 1992 to 2002 the headship rate for the overall population aged 25 to 34 increased by more than for any other age group. Cohort Effects The population cohort born during 1968 to 1977 had a headship rate of 49.1 percent at ages 25 to 34 in 2002 about 2 percentage points higher than the cohort born during 1958 to 1967 did at that age in 1992. That suggests that the headship rate among those 35 to 44 will increase, since the 1968-1977 cohort is entering that age bracket with headship rates higher than the previous cohort exhibited as they were entering their late 30s. The large doubled-up recent immigrant population aged 25 to 34 adds to the potential for an increase in headship rates for the 35 to 44 age group in the years ahead. Comparisons of headship rates for older ages, however, suggest that headship rates could be subject to downward pressure because the new entrants into those age brackets have established lower headship rates than those who preceded them. For example, the headship rate for the 35 to 44 year age bracket was 54.3 in 2002, when it was populated by those born from 1958 to 1967. Ten years earlier, when it was people born 1948 to 1957 that were in that age group, the headship rate (after adjusting for the effects of revisions in the survey in 1993 and 2001) was 55.4 percent. The share of the 35 to 44 year old population living with parents or other relatives actually increased between 1992 and 2002, with only modest declines since a peak in 1999. Increases in the number of householders living alone or as single parents failed to offset the decline in married-couple households. The 35 to 44 year old group includes a fairly large foreign-born share, with the associated potential to unbundle into additional households, but unlike the foreign-born population aged 25 to 34, the majority are not recent immigrants (only 40 percent arrived after 1989), and they may be more firmly set in their ways. Thus the late-baby boom cohort that is now entering their late forties starts off with relatively fewer household heads than the early boomers did when they reached that age, suggesting a (further) decline in headship rates for ages 45 to 54. The 55 to 64 year old and 65 plus age groups are subject to similar influences. Implications The shift in behavior patterns over the past decade have made no net contribution to growth in the number of households, beyond what is attributable to growth in the adult population and changes in the age structure, but the character of housing demand has been affected. Although the 1946-1964 Baby Boom is still the 800-pound gorilla dominating housing demand, the demands of Gen-Xers, Echo Boomers, immigrants, and minorities, will continue to represent a growing influence on housing demand and construction activity. Michael Carliner is a staff vice president with NAHB s Economics Group. For additional information, he can be reached by email at mcarliner@nahb.com. 1 If the average of all the monthly CPS data are used, rather than the March annual demographic supplement, the annual changes average 1.13 million from 1990 to 1996 and 1.26 million from 1996 to 2002, indicating a more modest acceleration. 2 The Census Bureau no longer uses the term head, having decided it was politically incorrect. In Census Bureau data and reports, the person (or one of the people) in whose name the housing unit is owned or rented is recorded as householder or reference person. 3 U.S. Census Bureau, Technical Assessment of A.C.E. Revision II (March 12, 2003). Earlier reports, issued in 2001, found a net population undercount in 2000 of 1.18 percent, with undercounts of 0.33 percent for occupied housing units and 3.37 percent for vacant units. In addition, there was an estimated net misclassification of 1.6 million vacant housing units as occupied. Given the large revision to the estimates of population undercount, those estimates of housing unit undercount and misclassification are suspect. The earlier estimates are in Report of the Executive Steering Committee for Accuracy and Coverage Evaluation Policy (March 1, 2001), and Census 2000 Housing Unit Coverage Study (ESCAP II Report 17, October 17, 2001) 4 Data from the 1990 and 2000 Censuses indicate that changes in headship rates reduced the average increase in the number of households by about 73,000 per year, due mainly to a decline in headship for ages 35 to 44. 5 Tavia Simmons and Martin O Connell, Married-Couple and Unmarried Partner Households: 2000 (U.S. Bureau of the Census, report CENSR-5, February 2003). This includes only those who indicated that an unmarried partner lived with the householder, rather than a roommate or other less intimate nonrelative co-resident. 6 For more on this subject, see Zhu Xiang Di, Yi Yang, and Xiaodong Lin, Young American Adults Living in Parental Homes (Harvard University Joint Center for Housing Studies, Working Paper W02-3, May 2002). 12 June 2003